At 01:07 PM 2/15/2003 +0100, Rik van Riel wrote:
>At 09:49 AM 2/15/2003 -0200, Rik van Riel wrote:
>>On Sat, 15 Feb 2003, Mike Galbraith wrote:
>>
>> > >Judging from your log, it ends up stalling kswapd and
>> > >dramatically increases the number of times that normal
>> > >processes need to go into the pageout code.
>> > >
>> > >If this provides an anti-thrashing benefit, something's
>> > >wrong with the VM in 2.5 ;)
>> >
>> > Which number are you looking at?
>>
>>pgscan 2751953 5328260 <== ? hmm
>>
>>kswapd_steal 380282 522126
>>pageoutrun 1107 1956
>>allocstall 3472 1238
>>
>>- we scan far less pages
>>- kswapd reclaims less pages
>>- we go into the pageout code less often
>>- allocations stall more often for a lack of free memory
>
>I would interpret that differently. I would say we scan less because we
>don't need to scan as much, kswapd reclaims less for the same reason, and
>ditto for pageout ;-) The reduction in scans does seem _way_ high though...
>
>wrt allocstall, I bet if I do a few runs, I'll see mucho variance there
>(could be wrong.. hunch)
After 10 consecutive runs, stalls for 2.5.61cfq2 were 11060 vs. 3656 for
2.5.61virgin.
(dang, hunch-o-meter needs calibration;)
-Mike
-
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@vger.kernel.org
More majordomo info at http://vger.kernel.org/majordomo-info.html
Please read the FAQ at http://www.tux.org/lkml/
This archive was generated by hypermail 2b29 : Sat Feb 15 2003 - 22:01:02 EST